Positive Text Reframing under Multi-strategy Optimization

TL;DR

The paper tackles positive text reframing, aiming to preserve original meaning while adding a complementary positive perspective. It introduces the Multi-Strategy Optimization Framework (MSOF), which combines reinforcement-based training with a positive sentiment and content-preservation rewards, decoding improvements, and a multi-dimensional re-ranking scheme to enforce strategy-consistency, content fidelity, and fluency. Extensive experiments on BART and T5 show significant gains in both unconstrained and controlled reframing tasks, across automatic metrics and human judgments, with the Top-k decoding variant often delivering the best overall performance. The work provides a practical, modular approach to controllable text reframing and introduces the RTQE metric to quantify reframing quality, contributing a robust toolchain for future research and applications in psychology-informed text generation.

Abstract

Differing from sentiment transfer, positive reframing seeks to substitute negative perspectives with positive expressions while preserving the original meaning. With the emergence of pre-trained language models (PLMs), it is possible to achieve acceptable results by fine-tuning PLMs. Nevertheless, generating fluent, diverse and task-constrained reframing text remains a significant challenge. To tackle this issue, a \textbf{m}ulti-\textbf{s}trategy \textbf{o}ptimization \textbf{f}ramework (MSOF) is proposed in this paper. Starting from the objective of positive reframing, we first design positive sentiment reward and content preservation reward to encourage the model to transform the negative expressions of the original text while ensuring the integrity and consistency of the semantics. Then, different decoding optimization approaches are introduced to improve the quality of text generation. Finally, based on the modeling formula of positive reframing, we propose a multi-dimensional re-ranking method that further selects candidate sentences from three dimensions: strategy consistency, text similarity and fluency. Extensive experiments on two Seq2Seq PLMs, BART and T5, demonstrate our framework achieves significant improvements on unconstrained and controlled positive reframing tasks.

Figures (5)

• Figure 1: The difference between sentiment transfer and positive reframing.
• Figure 2: The overall architecture of MSOF. We respectively use BART and T5 as the basic model for positive reframing. The positive sentiment reward and content preservation reward are applied to optimize the model training process. Then, we adopt various decoding improvement approaches (e.g. beam search, random sampling) during the decoding stage to improve the quality of text generation. Finally, multi-dimensional re-ranking is used to comprehensively evaluate candidate sentences and select the candidate with the highest score as the final output.
• Figure 3: The reinforcement training procedure of the Seq2Seq-based model.
• Figure 4: The overall procedure of reframe strategy classification.
• Figure 5: The model for RTQE.